Summary of Zipfian Whitening, by Sho Yokoi et al.
Zipfian Whitening
by Sho Yokoi, Han Bao, Hiroto Kurita, Hidetoshi Shimodaira
First submitted to arxiv on: 1 Nov 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Machine Learning (cs.LG); Machine Learning (stat.ML)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper highlights a significant issue with neural models’ word embedding spaces being skewed. The authors argue that most approaches for modeling and correcting this skewness assume uniform word frequencies, whereas in reality, word frequencies follow Zipf’s law, a highly non-uniform distribution. To address this, the authors propose a simple yet effective approach: perform PCA whitening weighted by empirical word frequency following Zipf’s law. This method outperforms established baselines and can be categorized theoretically as using an exponential family with a Zipfian base measure. The paper also discusses how popular NLP methods work well because they encode empirical word frequency into their probabilistic models. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This study shows that neural models’ word embedding spaces are not balanced, which affects task performance. To fix this, the researchers suggest using a new approach to correct the skewness. This method is surprisingly effective and works better than previous methods. The authors also explain why certain popular NLP techniques work well. |
Keywords
» Artificial intelligence » Embedding » Nlp » Pca
